What is Agentic AI for Banking with Claude Cowork?

It is a 2-day intensive training for banking professionals to learn how to delegate multi-step credit analysis tasks to AI agents using Claude Cowork.

When is the training schedule for 2026?

The training will be held on 19–20 October, 16–17 November, and 14–15 December 2026.

How much is the investment fee?

Rp.5.950.000 per participant, with an early bird fee of Rp.5.355.000 if paid 7 days before the event.

Agentic AI for Banking with Claude Cowork – Delegating Credit Analysis to AI Agents

Agentic AI Banking training – workspace with laptop, charts and calculator

Rate this post

Agentic AI Banking is the next step after chatbots and structured prompting: instead of asking an AI for one answer at a time, credit teams delegate whole pieces of work and review the result. Agentic AI for Banking with Claude Cowork – Delegating Credit Analysis to AI Agents is a 2-day intensive Claude Cowork training for corporate, commercial and SME credit professionals, held in Jakarta Selatan throughout 2026.

Course Description

From writing better prompts to directing a team of AI agents. This program is designed for bankers who already use AI chat and structured prompting (such as the 3-Tier Prompt Framework). It moves participants up the AI maturity ladder – from asking Claude for answers to delegating complete, multi-step banking work to it. It is a practical Banking AI Workshop for anyone who prepares, reviews or oversees credit proposals.

Day 1 (Level 3): build a single AI agent that plans, uses tools and files, and verifies its own output – applied to financial spreading, projection modeling and ratio analysis of a corporate borrower. This is hands-on AI credit analysis, not slideware.
Day 2 (Level 4): orchestrate a crew of specialist Cowork sub-agents – spreading, modeling, industry research, risk rating, memo writing and adversarial review – into an end-to-end credit workflow, with human approval gates and bank-grade governance.

Everything is hands-on in Claude Cowork, using PT Nusantara Prima Food Tbk (NPFO) as a common case borrower. Participants from different banks therefore work on the same case, and no institution’s confidential data enters the lab.

L1–L2 – where participants are now: chatbot plus structured prompts. One question, one answer, and the human does every step.
L3 – Day 1, Agentic AI: one agent plans, executes and self-checks a multi-step task with tools and files.
L4 – Day 2, multi-agent: a lead agent directs specialist sub-agents, and humans approve at defined gates.

To learn more about the tool itself, see Anthropic’s official Claude Cowork documentation. For industry context on why lenders are moving in this direction, read Moody’s overview of agentic AI in financial services.

Workshop Leader

Purwadi Nitimidjojo started his financial modelling career in early 2005, after working for the Indonesian Bank Restructuring Agency (IBRA / BPPN) during the monetary crisis from 1998 through 2004. At IBRA’s Credit Risk Management Division he developed various financial models for debt restructuring and corporate restructuring. He also saw that most of IBRA’s large debtors hired foreign financial consultants to build financial models, because local resources and financial modelling training courses were not yet available. That gap became his opportunity to deliver financial modelling training and develop best-practice models for clients across industries in Jakarta.

In early 2005 Purwadi teamed up with two friends, using “Edward, Farral & Peterson” and “Edward & Peterson” as their brands to run monthly public training courses and financial modelling services until early 2012, when he joined Deloitte Financial Advisory as a Financial Modelling Director. Over 10 years at Deloitte, he and his team developed best-practice financial models for clients in many industries and delivered monthly financial modelling and Power BI training for the public, clients and Deloitte’s internal staff in Indonesia, Singapore, Malaysia, Thailand and Cambodia.

Today he works as a financial modelling subcontractor for BDO and Deloitte, as well as an independent financial modelling consultant and trainer. His other programs include Financial Modeling & Analysis with Claude AI and Modeling Under Uncertainty with Claude.

Training Schedule 2026

Full-day training, 9am–5pm, over 2 days. Available batches for 2026:

19–20 October 2026
16–17 November 2026
14–15 December 2026

Venue Training

The Agentic AI Banking training is held at one of the following hotels in Jakarta Selatan (the final venue is confirmed before the event):

Swiss-Belinn Simatupang, Jakarta Selatan
Swiss-Belhotel Pondok Indah, Jakarta Selatan
Ibis Styles Simatupang, Cilandak, Jakarta Selatan
Hotel TMG Tebet, Tebet, Jakarta Selatan

Investment Fee

Rp.5.950.000 / participant
Early Bird Rp.5.355.000 / participant (paid 7 days before the event, H-7)

The investment fee includes:

Training Material
Certificate
Lunch & Coffee Break
Souvenir
Door Prize

This program can also be delivered as an In-House Training with a maximum of 20 participants per batch.

Course Objectives

By the end of this program, participants will be able to:

Explain the agent: distinguish chatbot, agent and multi-agent systems, and map each to the Engineering Stack: prompt → context → harness → loop.
Engineer the context: package bank credit policies, templates and procedures into reusable Cowork skills, project instructions and connectors.
Delegate multi-step work: direct a single agent to spread statements, build a projection model and compute credit ratios (DSCR, DuPont ROE).
Build verification loops: design plan-execute-verify cycles, reconciliation checks and the Three-Pass review so errors surface instead of hiding.
Orchestrate a crew: design Cowork sub-agent crews for pipeline, fan-out and reviewer work, with role cards and handoff contracts.
Govern responsibly: place human approval gates, protect confidential data (UU PDP) and align agent use with model-risk and OJK expectations.

Who Should Attend

Corporate, commercial and SME credit analysts
Relationship managers who prepare credit proposals
Credit risk, credit review and portfolio monitoring officers
Digital transformation, AI and innovation teams in banking
Model risk, compliance and internal audit staff overseeing AI use
Banks and multi-finance companies of any size – individual or group registration

Prerequisites: comfortable with everyday AI chat and structured prompting; working knowledge of financial statements and credit analysis; your own laptop with the Claude Desktop app and Cowork on a paid Claude plan (Pro, Max, Team or Enterprise). Team and Enterprise users need admin approval for the plugins and connectors used in the labs. No bank or customer data is needed – all labs use the NPFO case.

Course Outline

Day 1 – Level 3, Agentic AI: One Agent, End-to-End Work

08:30 – 09:00, Registration & Opening: participant introductions, 3-Tier Prompt Framework refresher, the L1 → L4 AI maturity ladder
09:00 – 10:15, Module 1 – Anatomy of an AI Agent: goal, plan, tools, memory, loop; chatbot vs agent; live Cowork demo
10:30 – 12:00, Module 2 – Context Engineering in Cowork: skills (SKILL.md), project instructions, connectors, folder access; build a bank credit-analysis skill from a model credit policy and template
13:00 – 14:45, Module 3 – Hands-On Agentic Credit Analysis: the agent ingests NPFO financials, spreads statements, builds a 3-statement projection, then runs DuPont, profitability, efficiency, liquidity, leverage, capital structure and debt service coverage analysis
15:00 – 16:15, Module 4 – Verify-and-Control Loops: plan-execute-verify, balance checks and reconciliations, Three-Pass Method, agent failure modes
16:15 – 17:00, Day 1 Clinic & Wrap-Up: map your own workflows to agent candidates; Q&A

Day 2 – Level 4, Multi-Agent Orchestration: Directing the Crew

08:30 – 09:00, Day 1 Recap: when one agent is no longer enough
09:00 – 10:15, Module 5 – Multi-Agent Architecture Patterns: lead agent and sub-agents; sequential pipeline, parallel fan-out, reviewer/critic; when not to go multi-agent
10:30 – 12:00, Module 6 – Designing the Credit Workflow Crew: role cards for Spreading, Modeling, Industry Research, Risk Rating, Memo Writer and Adversarial Reviewer, plus handoff contracts
13:00 – 14:45, Module 7 – Capstone, Orchestrate the Credit Memo: team build of an end-to-end NPFO credit proposal with Cowork sub-agents, packaged as a plugin, with human approval gates
15:00 – 16:00, Module 8 – Governance, Risk & Scale: human-in-the-loop gates, data confidentiality and UU PDP, audit trail, model-risk and OJK alignment, rollout roadmap
16:00 – 17:00, Capstone Showcase & Closing: team presentations with adversarial critique, action plans, certificates

Key Takeaways

Agent Mindset: stop asking for answers – delegate outcomes, then review the work.
Reusable Skills: a credit-analysis Cowork skill template and project setup, ready to adapt to your bank’s policies.
Verified Output: checks, reconciliations and Three-Pass review built into every run.
Crew Blueprint: a six-role Cowork plugin for the credit workflow, with handoff contracts.
Human Approval Gate: clear approval points – AI prepares, the banker decides and signs off.
Adoption Roadmap: a prioritized list of your institution’s workflows to pilot at Level 3 and Level 4.

This program is part of the training services of Seventh Grace Consulting, a public and in-house training provider operating since 2009. Looking for more options in finance and modeling? Browse the complete Seventh training catalog or the earlier Business Plan Modeling with Claude Cowork and Claude Code program.